Thomas Leonard

I've rebased it and squashed (so there are no commits in the history with failing tests). I slightly modified the test to check that the parent cancellation takes effect on...

I've seen that bug locally, but not in CI as far as I can recall: https://github.com/ocaml-multicore/eio/issues/700 https://bugs.launchpad.net/ubuntu/+source/glibc/+bug/1899800/comments/5 says: > This bug was fixed in the package glibc - 2.32-0ubuntu5

That is surprising. It works on my machine. And it doesn't call `writev` at all (according to strace). Do you have uring blocked somehow? Maybe you're running in a Docker...

Actually, I guess this could happen if the uring writev job returned EINTR. Though I don't see why it would do that.

Does it happen with fewer domains? Assuming it is uring (strace will say), you should be able to trace the completions with a `trace.bt` like this: ``` BEGIN { @op[IORING_OP_NOP]...

Thanks @patricoferris - with your code it reproduces for me. I asked about it at https://github.com/axboe/liburing/discussions/1452 and it seems this might be a uring problem. I'm adding support for setup...

This seems not quite the same as your #336 proposal - in fact, it seems incompatible with that idea (of storing the protected resource inside the mutex). Here it's passed...

The argument to fallback is ``[`Msg of string]``.

The skip feature needs adding to MDX itself (not Eio). But existing md files will need to be modified to use it.

You'll need to add a test to `test/bin` I would expect.